"Growth Models" Alone are Insufficient NCLB Reform

[Bob Schaeffer <bobschaeffer@earthlink.net>]

GROUP LOBBIES FOR TEST OPTIONS: DEL. EDUCATORS WANT ADAPTIVE SYSTEMS USED

      (Delaware) News Journal -- November 5, 2007
                  by Alison Kepner

Worried that current state tests mismeasure how many students are doing, 
a group of Delaware educators and community leaders is asking Congress 
to allow multigrade, computer-adaptive tests to determine student 
progress under the federal No Child Left Behind act.

Two members of Congress introduced a bill last week that would give 
states the option to use adaptive testing.

A Delaware group-commissioned study, released last week, found such 
tests documented progress that the Delaware Student Testing Program 
missed, especially among low-income, minority or 
English-language-learner students.

By the 2013-14 school year, all students must be proficient in reading, 
writing and math for a school to meet federal accountability goals. The 
2001 federal law originally required states to determine proficiency 
using pass-fail tests that measure whether students perform at their 
grade level.

Last year, Delaware became one of 10 states to pilot an alternate model 
that considers individual students' scores over time, giving schools 
credit for improvement even if students had not yet reached proficiency.

But the Delaware group argues that the pilot approach, while an 
improvement, is still flawed because it is based on single-grade-level 
exams. To show true growth -- particularly among students performing far 
above or below grade level -- tests must determine not just whether a 
student is performing at his grade level but, more specifically, at 
which grade level he is performing, the group said.

In 2005, the Delaware Statewide Academic Growth Assessment Pilot 
steering committee -- whose members include superintendents, nonprofit 
education foundation leaders and state Education Secretary Valerie 
Woodruff -- began testing computer-adaptive tests in four dozen public 
schools across Delaware.

This summer, the group commissioned a Tennessee research center to 
analyze the results, comparing Delaware Student Testing Program and 
adaptive-test scores. The study found a majority of students who failed 
the DSTP are in fact improving academically. That means many schools are 
making progress but not getting federal credit for it.

The impact was greatest in schools with the highest percentages of 
low-income, minority and English-language-learner students: Using the 
adaptive test, almost 60 percent of high-poverty Delaware schools showed 
student improvement, compared to 25 percent that did so under the 
current models.

The computerized adaptive tests adjust difficulty based on how a student 
answers. If a fifth-grader struggles with fifth-grade questions, the 
test drops to easier ones to find whether the child is at, for example, 
a second- or fourth-grade level. Likewise, if the student excels, the 
difficulty increases to determine at what higher grade the student is 
performing.

The Delaware committee -- with support from teacher, administrator and 
business groups -- is asking Congress to give states the option of using 
such tests.

"It makes sense, and it will give us a more meaningful reflection of how 
our schools are doing," Appoquinimink Superintendent Tony Marchio said.

Although his district wasn't part of the pilot program, Marchio wants to 
bring computer-adaptive testing to his schools.

"It helps us to identify exactly where the students are when we get 
them," he said. "Even if they come to us below grade level ... are they 
on the path to proficiency? That's not reflected in the current system."

The group already has the backing of Sen. Tom Carper, D-Del., who said 
"we need to allow states to explore the benefits of adaptive testing -- 
which include faster results and more accurate information on each 
student's achievement over time."

Woodruff, though, worries about giving adaptive tests too much weight. 
While she agrees they can be an effective tool, she differs with 
committee members on how that tool should be used.

"An adaptive assessment, in terms of figuring out where kids are 
throughout the year, makes some sense, but when it comes to the point of 
whether or not a student has or has not met proficiency, then I think 
the adaptive is a stretch," she said.

"If you have a child who is in the eighth grade and is supposed to be 
functioning at the eighth-grade level and that student is not able to do 
that, then by doing an adaptive assessment, what you are doing is 
essentially not really measuring where that student is supposed to be 
and the standard to which the school is supposed to be providing."

While she thinks the Delaware group members' motives are good, she 
worries about potential consequences of such a change.

"Some would see it as an attempt to game the system and really not hold 
schools accountable to make students meet the measure," she said.

As NCLB, now up for reauthorization, is retooled, Woodruff wants 
legislators to give states the ability to use multiple measures to 
decide whether a school is making progress. That may mean looking at 
adaptive test results in combination with scores from a status test such 
as DSTP.

"I see it as a valuable measure, but I do not see it as the measure," 
Woodruff said.

But Nancy Doorey, pilot coordinator for the steering committee, sees a 
bigger danger in not making the change.

By not recognizing student growth, "we've created a huge incentive for 
states to lower their standards," she said. "If we can accurately 
measure growth, the states will have more courage to raise their standards."

  
http://www.delawareonline.com/apps/pbcs.dll/article?AID=/20071105/NEWS/711050377/1006/NEWS